Sorry for the late reply Maru, didn't see your answer before.
I'll try to explain my workflow.
I usually setup my architectural visualisation scenes in the way that each camera view is on it's own frame (cam1 on frame 1, cam2 on frame 2...)
In each frame I set sun position, maybe move a bit some objects in scene if needed to make composition better, add some objects for one frame and move it out of the other...
All of this is animated so that I can batch render my project over night.
This method works great for me until I need to add for example people or cars with motion blur, because I also get motion blur on trees or some other object that i animated to fit better in composition.
So reason why I would need this feature is probably not so common, maybe I should think about some other workflow for this (probably using scene states and turn on/off layers or something like that)
Thanks!